2.1. Src and the SH3 Domain
Among a number of proteins involved in eukaryotic cell signaling, there is a common sequence motif called the SH3 domain. It is 50-70 amino acids in length, moderately conserved in primary structure, and can be present from one to several times in a large number of proteins involved in signal transduction and in cytoskeletal proteins.
The protein pp60c-src represents a family of at least nine non-receptor protein tyrosine kinases (NR-PTKs). Members of this family share an overall structural organization comprising a series of catalytic and non-catalytic domains. In Src, a 14-amino-acid myristylation signal resides at the extreme amino-terminus, and is followed by a unique region that is not highly conserved among family members. Following this region are two highly conserved 60- and 100-amino-acid regions, the Src homology (SH) domains 3 and 2, respectively. SH2 and SH3 domains have been shown to play an important role in mediating protein-protein interactions in a variety of signaling pathways. Koch, C. A., et al., in Science (1991) 252:668-74. The carboxy-terminal half of Src contains the PTK catalytic domain, as well as a negative regulatory tyrosine (Y527) near the carboxy terminus. Phosphorylation of this residue (e.g., by Csk) results in the inhibition of PTK activity. Cooper, J. A., et al., in Science (1986) 231:1431-1434. Mutation of Y527.fwdarw.F generates forms of Src with increased PTK and oncogenic activity. Cartwright, C. A., et al., in Cell (1987) 49:83-91; Kmiecik, T. E., et al., in Cell (1987) 49:65-73; and Piwicna-Worms, H., et al., in Cell (1987) 75-82.
The fact that some mutations which result in increased Src PTK and transforming activity map to the Src SH2 (Seidel-Dugan, C., et al., in Mol. Cell. Biol. (1992) 12:1835-45; and Hirai, H. and Varmus, H. E. in Mol. Cell. Biol. (1990) 10:1307-1318) and SH3 domains (Seidel-Dugan, C., et al., supra; Hirai, H. and Varmus, H. E., supra; Superti-Furga, G., et al., in Embo. J. (1993) 12:2625-34; and Potts, W. M., et al., in Oncogene Res. (1988) 3:343-355) suggests a negative regulatory role for these domains. That phosphotyrosine residues within specific sequence contexts represent high affinity ligands for SH2 domains suggests a model in which the SH2 domain participates in Y527-mediated inhibition of PTK activity by binding phosphorylated Y527, thereby locking the kinase domain in an inactive configuration. Matsuda, M., Mayer, B. J., et al., in Science (1990) 248:1537-1539. This model is supported by the observation that phosphopeptides corresponding to the carboxy-terminal tail of Src bind active, but not inactive, variants of Src. Roussel, R. R., et al., in Proc. Natl. Acad. Sci. U S A (1991) 88:10696-700; and Liu, X., et al., in Oncogene (1993) 8:1119-1126.
The mechanism of SH3-mediated inhibition of Src PTK activity remains unclear. There is evidence that pY527-mediated inhibition of Src PTK activity involves the SH3 domain as well as the SH2 domain. Okada, M., Howell, et al., in J. Biol. Chem. (1993) 268:18070-5; Murphy, S. M., et al., in Mol. Cell. Biol. (1993) 13:5290-300; and Superti-Furga, G., et al., supra. Although these effects are thought to be a consequence of SH3-mediated protein-protein interactions, precisely how the Src SH3 domain exerts its negative regulatory effect is unclear. Identification of high affinity ligands for the Src SH3 domain could help resolve these issues.
2.2. Protein Tyrosine Kinases and the Immune Response
Src-related tyrosine kinases are expressed in a variety of cell types including those of the immune system (lymphocytes, T cells, B cells, and natural killer cells) and the central nervous system (neural cells, neurons, oligodendrocytes, parts of the cerebellum, and the like). Umemori, H. et al., in Brain Res. Mol. Brain Res. (1992) Dec. 16(3-4):303-310. Their presence in these cells and tissues and their interaction with specific cell surface receptors and immunomodulatory proteins (such as T cell antigen receptor, CD14, CD2, CD4, CD40 or CD45) suggest that these kinases serve an important role in the signalling pathways of not only the central nervous system but of the immune system, as well. See, e.g., Ren, C. L. et al., in J. Exp. Med. (1994) 179(2):673-680 (signal transduction via CD40 involves activation of Lyn kinase); Donovan, J. A. and Koretzky, G. A., in J. Am. Soc. Nephrol. (1993) 4(4):976-985 (CD45, the immune response, and regulation of Lck and Fyn kinases); and Carmo, A. M. et al., in Eur. J. Immunol. (1993) 23(9):2196-2201 (physical association of the cytoplasmic domain of CD2 with p56lck and p59fyn).
For instance, mice with disruptions in their Src-like genes, Hck and Fgr, possess macrophages with impaired phagocytic activity or exhibit a novel immunodeficiency characterized by an increased susceptibility to infection with Listeria monocytogenes. Lowell, C. A. et al., in Genes Dev. (1994) 8(4):387-398. Also, it has been shown that bacterial lipopolysaccharide (LPS) activates CD14-associated p56lyn, p68hck, and p59c-fgr, while inducing the production of lymphokines, such as TNF-alpha, IL-1, IL-6, and IL-8. Inhibition of the protein tyrosine kinases blocks production of TNF-alpha and IL-1.
2.3. SH3 Binding Peptides
As mentioned above, it has long been suspected that SH3 domains are sites of protein-protein interaction, but it has been unclear what SH3 domains actually bind. Efforts to identify ligands for SH3 domains have led to the characterization of a number of SH3-binding proteins, including 3BP1 and 2 (Ren, R., Mayer, et al., in Science (1993) 259:1157-61), SOS (Olivier, J. P., et al., in Cell (1993) 73:179-91; and Rozakis-Adcock, M., et al., in Nature (1993) 363:83-5), p85 PI-3' Kinase (Xingquan, L., et al., in Mol. Cell. Biol. (1993) 13:5225-5232), dynamin (Gout, I., et al., in Cell (1993) 75:25-36), AFAP-110 (Flynn, D. C., et al., in Mol. Cell. Biol. (1993) 13:7892-7900), and CD42 (Barfod, E. T., et al., in J. Biol. Chem. (1993) 268:26059-26062). These proteins tend to possess short, proline-rich stretches of amino acids, some of which have been directly implicated in SH3 binding. A variety of consensus sequences have been proposed, although the similarity among proline-rich regions of different SH3-binding proteins tends to be fairly low. Also, attempts to build consensus sequences are likely complicated by the incorporation of data from proteins that bind different SH3 domains.
Thus, Cicchetti, P., et al., in Science (1992) 257:803-806, published their work relating to the isolation and sequencing of two naturally-occurring proteins that could be bound in vitro by the SH3 domain of the abl oncogene product. These workers found that SH3 domains bind short, proline-rich regions of such proteins. Subsequently, this same group disclosed further results (Ren, R. et al., supra) in which the SH3 binding sites of the SH3 binding proteins were localized to "a nine- or ten-amino acid stretch rich in proline residues." A consensus sequence incorporating the features of the SH3 binding sites of four SH3 binding proteins was proposed: XPXXPPP.Yen.XP (SEQ ID NO:1), wherein X indicates a position in the amino acid sequence which is not conserved among the four SH3 binding proteins, P represents proline, and .Yen. indicates a hydrophobic amino acid residue, such as P or L.
The screening of complex random peptide libraries has been used to identify peptide epitopes for monoclonal (Scott, J. K. and Smith, G. P. in Science (1990) 249:386-390) and polyclonal (Kay, B. K., et al., in Gene (1993) 128:59-65) antibodies, as well as peptide ligands for a variety of proteins, including streptavidin (Devlin, J. J., et al., in Science (1990) 249:404-406; and Lam, K., et al., in Nature (1991) 354:82-84), the endoplasmic reticulum chaperone BiP (Blond-Elguindi, S., et al., in Cell (1993) 75:717-728), and CaM (Dedman, J. R., et al., in J. Biol. Chem. (1993) 268:23025-23030).
Recently, Chen, J. K. et al., in J. Am. Chem. Soc. (1993) 115:12591-12592, described ligands for the SH3 domain of phosphatidylinositol 3-kinase (PI-3' Kinase) which were isolated from a biased combinatorial library. A "biased" library is to be distinguished from a "random" library in that the amino acid residue at certain positions of the synthetic peptide are fixed, i.e., not allowed to vary in a random fashion. Indeed, as stated by these research workers, screening of a "random" combinatorial library failed to yield suitable ligands for a PI-3' Kinase SH3 domain probe. The binding affinities of these unsuitable ligands was described as weak, &gt;100 .mu.M, based on dissociation constants measured by the Biosensor System (BIAcore).
More recently, Yu, et al. (Yu, H., et al., in Cell (1994) 76:933-945) used a "biased" synthetic peptide library of the form XXXPPXPXX (SEQ ID NO:2), wherein X represents any amino acid other than cysteine, to identify a series of peptides which bind the Src and PI-3' Kinase SH3 domains. The bias was accomplished by fixing the proline residues at the specific amino acid positions indicated for the "random" peptide. As stated previously, without this bias, the technique disclosed fails to identify any SH3 domain-binding peptides.
A consensus sequence, based on 13 binding peptides was suggested: RXLPPRPXX (SEQ ID NO:3), where X tends to be a basic residue (like R, K or H). The binding affinities of several SH3 binding peptides were disclosed as ranging from 8.7 to 30 .mu.M. A "composite" peptide, RKLPPRPRR (SEQ ID NO:4), was reported to have a binding affinity of 7.6 .mu.M. This value compares favorably to the binding affinity of the peptide, VPPPVPPRRR (SEQ ID NO:5), to the N-terminal SH3 domain of Grb2. See, Kraulis, P. J. J. Appl. Crystallogr. (1991) 24:946. Recognizing the limitations of their technique, Chen and co-workers, supra, stated that their results "illustrate the utility of biased combinatorial libraries for ligand discovery in systems where there is some general knowledge of the ligand-binding characteristics of the receptor" (emphasis added).
Yu and co-workers, supra, further described an SH3 binding site consensus sequence, Xp.O slashed.PpXP (SEQ ID NO:6), wherein X represents non-conserved residues, .O slashed. represents hydrophobic residues, P is proline, and p represents residues that tend to be proline. A consensus motif of RXLPPRPXX (SEQ ID NO:7), where X represents any amino acid other than cysteine, was proposed for ligands of PI-3' Kinase SH3 domain. A consensus motif of RXLPPLPR.phi. (SEQ ID NO:8), where .phi. represents hydrophobic residues, was proposed for ligands of Src SH3 domain. Still, the dissociation constants reported for the 9-mer peptides ranged only from about 8-70 .mu.M and selectivity between one type of SH3 domain and another was relatively poor, the K.sub.D s differing by only about a factor of four.
Hence, there remains a need to develop techniques for the identification of Src SH3 binding peptides which do not rely on such "biased" combinatorial peptide libraries that are limited to a partially predetermined set of amino acid sequences. Indeed, the isolation of SH3 binding peptides from a "random" peptide library has not been achieved successfully before now. Furthermore, particular peptides having much greater binding affinities, whether general or more selective binding for specific SH3 domains, remain to be identified. Binding peptides specific for particular SH3 domains are useful, for example, in modulating the activity of a particular SH3 domain-containing protein, while leaving others bearing an SH3 domain unaffected. Still, the more promiscuous general binding peptides are useful for the modulation of a broad spectrum of SH3 domain-containing proteins.
The present invention relates to such SH3 binding peptides, methods for their identification, and compositions comprising same. In particular, peptides comprising particular sequences of amino acid residues are disclosed which were isolated from random peptide libraries. In the present invention, clones were isolated from a phage-displayed random peptide library which exhibited strong binding affinities for SH3 domain-containing protein targets. Some of these protein targets, include Abl, Src, Grb2, PLC-.delta., PLC-.gamma., Ras GAP, Nck, and p85 PI-3' Kinase. From the nucleotide sequence of the binding phage, the amino acid sequence of the peptide inserts has been deduced. Synthetic peptides having the desired amino acid sequences are shown to bind the SH3 domain of the target proteins. In particular, synthetic peptides combining a core consensus sequence and additional amino acid residues flanking the core sequence are especially effective at binding to particular target protein SH3 domains. The SH3 binding peptides disclosed herein can be utilized in a number of ways, including the potential modulation of oncogenic protein activity in vivo. These peptides also serve as useful leads in the production of peptidomimetic drugs that modulate a large class of proteins involved in signal transduction pathways and oncogenesis.